Inverter Firmware v4.12.R

[johu]

After some experiments with a Swedish user who actually runs a Nissan Leaf motor at high voltage I have updated the velocity calculation in resolver and sin/cos mode. It now waits for the resolver signal to become stable and it resets the calculated frequency to 0 if there is very little change for a period of time.

If you're using resolver or sin/cos feedback I very much recommend upgrading to this firmware. For everyone else it is not relevant.

https://github.com/jsphuebner/stm32-sin ... ag/v4.12.R

[PRB617]

I got a couple of Axpert 5MKS inverters which I have replaced with new one's for customers which gave me a Error 51. The inverters still work but will come up with the error intermittently.

[damian.lo]

Hello,

Johannes can You explain what can be wrong if with firmware version 3.93 my inverter with resolver is working well, but after upgrading firmware to this 4.12 version it's not starting? When I wrote "start 2" i hear only little click in Leaf motor but nothing more. When I tried a little rotate axis by hand it little holds - thats all.
Then I back to 3.93 and works well - what is new in this firmware (else than sin/cos calculating)? When I put plot for angle on 3.93 it seems good, so can be more?

[johu]

Hi Damian,

thanks for pointing it out, I'll research that.

[johu]

I think I found it. In version 4.75 (uh, what happened to my version counting? Anyway, 4.75 is older than 4.12.R) I swapped sin/cos channels to mach the cable loom of the Leaf inverter. I think there is your problem. Can you swap sin/cos and try again?

[arber333]

I have Installed 4.12r in my Mazda inverter and at first all was good. But today I had to drive through heavy traffic and there were severe vibrations below 2k rpm. I used the same settings than 4.71 version you sent me for testing last year.
What did you change to lessen vibration on ACIM motors at low rpm?

[damian.lo]

Hi,
Of course I will look at this and try. Probably tomorrow, because today my motor is fitting into car to make a flange.
Stupid question: swap sin/cos between resolver interface and brain board?

[arber333]

Yesterday i installed 4.12R in my Mazda with ACIM motor and AB encoder and tried to drive. I noticed nothing special when starting etc... However in the dense traffic i noticed a lot of vibration at transition from drive to regen. This was much pronounced when throttle was held at small rpm. Amplitude kept rising and i had to use the brake to stop the shakes.

Today i replaced it with 4.71 version. It supposed to have additional encoder filtering. And it works! Driveoff is noticably smoother and transition at low rpm is felt like a small kick not lots of vibrations. It is obvious better PID is applied here since initially at start throttle applies larger torque that is lessened if the throttle is held. This feels very natural and smooth. Johannes why not apply this math to the later code?
I am attaching 4.71 code here, you may try it yourself and decide.

[johu]

Yes that version had the position tracker. It seemed to produce much more vibration on regen (you also reported that) so I ended up removing it. Can you test regen again (especially strong regen)

[arber333]

Yes that version had the position tracker. It seemed to produce much more vibration on regen (you also reported that) so I ended up removing it. Can you test regen again (especially strong regen)

Yes yes.... regen at full power is still teeth rattling event, however i reduced regen effect enough to be comfortable. I think driving smoothness is what makes a good experience in day to day driving. Regen can also be shutoff in snow or rain etc... I am not sure, i guess my motor just behaves better with higher startup slip. Strange thing is that smooth drive with that slip is possible with positive throttle but regen causes violent shaking in transmission coupling...

[damian.lo]

Ok, so like you Johannes said I tried a moment ago to change sin/cos signals going to brain board - nothing changes. Then I tried to change phase 1 and phase 2 inverted - the same.
What I found - was necessary to change udcgain/udcofs because udc changes with new firmware version.
I'm attaching parameters file - maybe You will found what is necessary to modify. Small question - on this firmware version is new parameter in encoder section: sin-cos (previously was only resolver) - for trial I tried also this - was no better. Also second thing - resolver polepairs - like You see I have =1, but I tried to change to 2 - the same.
What can be else?

[johu]

I think now I see what happened. For Leaf motor respolepairs must be 4, also polepairs. sin/cos is for hall effect chips that work with a permanent magnet. They deliver the same sin/cos signal like a resolver but without excitation.
Thanks for testing this! I am so down on parameter documentation...

[damian.lo]

Ok, now is working and motor is spinning but another thing - from time to time motor won't start and I have overcurrent limit and sometimes a small changing angle by hand before start is helping to start. I don't know why because on 3.93 I hadn't this effect.
Is some difference between "Start 2" and pushing "start in man. mode" button?
For "Start 2" is important fslipmin/fslipmax value? And last for now - still my current reading is not stable. I don't have any idea why is going from ex. -8 to 8A is it normal?

[johu]

"start 2" and pushing the button does the same. fslipmin/max is not relevant in sync mode.
I'm trying to recall, are you using the Melexis sensors?

[arber333]

Your offset is not good enough.
Just today i managed to turn the motor again.
I managed to find good offset when i get smooth rotation and high rpm. If i try to go further rotation gets rougher and there is a chance to throw OClimit.
It seems boost is some factor with enabling rpm to extend higher. If i use boost 12000 i can get to 6000rpm at 400Hz vs 10000 boost. Weakening is set 600Hz and supposedly ineffective. But really i notice that offset is causing some weakening to happen and thus motor is spinning faster while consuming more amps.
If i use offset in the other way direction is reversed. It works! Well i had offset tuned to 32000 and then i changed to 48000 and it worked. Motor is turning the other way smoothly.
I need to put some coolant into inverter. After some minutes of running IGBT module reached 69°C and it tripped overtemp error.

[damian.lo]

Ok, so I will try to find better offset. Now I have 9000, what I remember above 10000 I had overcurrent immediatelly. So probably will be in next 180 degrees. After 5 minutes on bench my temp. goes from 19 to 23 deg.C
What I really is mysterious for me why on old 3.93 was good and on 4.12 are some problems. Arber, Your current diagram looks similar like mine?

[arber333]

I had some ups and downs... check here viewtopic.php?f=2&t=15&start=40

[damian.lo]

Yes, You have ups and downs but there are always about one value, let's say 10+/- 3. Mine is 2 +/- 6, strange
From beginning of my inverter I have not stable current read. Even Johannes gave ideas if I have some lamps/magnets/other interferences to Melexis chips but I haven't anything like this near to chip. Cables (strip) from Melexis 1 and 2 should be twisted and in shield maybe? I used normal tape of 3 wires. Probably at the end I will must to use Tamura current sensor.

[nailgg]

Yes, You have ups and downs but there are always about one value, let's say 10+/- 3. Mine is 2 +/- 6, strange
From beginning of my inverter I have not stable current read. Even Johannes gave ideas if I have some lamps/magnets/other interferences to Melexis chips but I haven't anything like this near to chip. Cables (strip) from Melexis 1 and 2 should be twisted and in shield maybe? I used normal tape of 3 wires. Probably at the end I will must to use Tamura current sensor.

Was the current graph you posted taken with no actual current on motor? Can you please elaborate the issue?

[damian.lo]

It was taken with motor running on bench with Start Inverter in Manual Mode.
Now I have 36V DCrail for trials.

[nailgg]

So sorry if I’m stating the obvious but you’re reading an AC signal there and it’s normal to see an alternating signal from -8 to +8 amps.

[damian.lo]

From the motor I know, but I don't know why I was thinking that after transformating by code on gauges or plot the result will be some absolute value, let's say about 8A (+/-). So was good reading, but I was expecting like RMS value?

[nailgg]

You should plot il1rms and il2rms for the RMS values. It seems like you plotted il1 and il2, these are the raw currents.

[damian.lo]

Exactly, I plotted il1 and il2 and expect values for rms. Thanks nailgg, that You kicked me in a.. to wake up
Good, one thing closed. Now I'm wodering why in v4.12 when I press button with last errors green led is still on - like blocked cpu. In this moment I need to reboot brain.

[arber333]

Yes, You have ups and downs but there are always about one value, let's say 10+/- 3. Mine is 2 +/- 6, strange
From beginning of my inverter I have not stable current read. Even Johannes gave ideas if I have some lamps/magnets/other interferences to Melexis chips but I haven't anything like this near to chip. Cables (strip) from Melexis 1 and 2 should be twisted and in shield maybe? I used normal tape of 3 wires. Probably at the end I will must to use Tamura current sensor.

I am now using stock Volt inverter. Really i am not designing another power stage... ever! The last one shortened my life substantially.
Huh you are using Melexis sensors? How close are the AC rails to each other? There is a lot of interference if you do not design the AC rails correct you know. After a friend designed and cut new copper tracks for me in such a way that they cover each other a lot of ringing was canceled. But still rails go through a loop. With Melexis sensors there is no loop, just sniffing the field. Can you use a handheld current clamp directly on the phase cable while you observe melexis output? Should give you a comparison.

See some photos of correct power stage. Aleš made a masterpiece design. And even better it works now a year without any problems.
(Johannes i cant seem to upload images, rar file is good though?)